Hybrid CNN-LBP Model for Realtime Emotion Classification

Facial emotion detection helps in understanding human behaviour by analyzing expressions. It is widely used in fields like healthcare, security, and education to enhance interactions and decision- making. Traditional methods rely on manual observation, which can be subjective and inefficient, making automated detection a more effective solution. This study proposes two methods: Method 1 uses Convolutional Neural Networks (CNNs) with data augmentation, while Method 2 integrates Local Binary Pattern (LBP) features with CNNs for enhanced feature extraction. Both methods are evaluated on the FER2013 dataset, focusing on five emotions: sad, angry, happy, neutral, and surprise. Enhancements include data augmentation for balanced data and real-time detection via live camera access. Method 2 shows improved accuracy due to LBP features, making the system more precise and reliable for applications in human-computer interaction, mental health analysis, and behavioral research.